[Pgpool-general] Error: pool_add_sent_message: portal "" already exists

[Jonathan Regeimbal]

Hi everyone,

I'm looking for help (and hopefully a solution) involving an error I am 
getting while using PGPool2 in master/slave +LB mode on top of PG9 in 
streaming replication mode.  In general Pgpool2 is working great, but as 
noted in the subject every so often we get portal errors back from 
PGPool2 that result in internal server errors.  When our application is 
connected directly to PG9, these errors do not occur.  Digging deeper 
these errors seem to commonly occur in PGPool2 while in Load Balance 
mode, but only sometimes are they reported back to DBI as an error.  
Here is an excerpt from Pgpool2's debug log.

May 13 21:13:22 ip-10-119-14-100 pgpool: 2011-05-13 21:13:22 DEBUG: pid 
9083: Execute: portal name <>
May 13 21:13:22 ip-10-119-14-100 pgpool: 2011-05-13 21:13:22 DEBUG: pid 
9080: Execute: portal name <>
May 13 21:13:22 ip-10-119-14-100 pgpool: 2011-05-13 21:13:22 DEBUG: pid 
9082: Execute: portal name <>
May 13 21:13:22 ip-10-119-14-100 pgpool: 2011-05-13 21:13:22 DEBUG: pid 
9085: pool_add_sent_message: portal "" already exists
May 13 21:13:22 ip-10-119-14-100 pgpool: 2011-05-13 21:13:22 DEBUG: pid 
9085: Execute: portal name <>
May 13 21:13:22 ip-10-119-14-100 pgpool: 2011-05-13 21:13:22 DEBUG: pid 
9078: pool_add_sent_message: portal "" already exists
May 13 21:13:22 ip-10-119-14-100 pgpool: 2011-05-13 21:13:22 DEBUG: pid 
9051: Execute: portal name <>
May 13 21:13:22 ip-10-119-14-100 pgpool: 2011-05-13 21:13:22 DEBUG: pid 
9078: Execute: portal name <>
May 13 21:13:22 ip-10-119-14-100 pgpool: 2011-05-13 21:13:22 DEBUG: pid 
9080: pool_add_sent_message: portal "" already exists
May 13 21:13:22 ip-10-119-14-100 pgpool: 2011-05-13 21:13:22 DEBUG: pid 
9080: Execute: portal name <>
May 13 21:13:22 ip-10-119-14-100 pgpool: 2011-05-13 21:13:22 DEBUG: pid 
8910: pool_add_sent_message: portal "" already exists

On occasion Pgpool2 logs this error at the LOG level in addition to the 
DEBUG level:

May 13 20:39:12 web2 pgpool: 2011-05-13 20:39:12 LOG:   pid 8107: 
pool_send_and_wait: Error or notice message from backend: : DB node id: 
0 backend pid: 8157 statement: P message: portal "" does not exist

And when that happens our application receives the error too (And 
generates an Internal server error)

This is the error as logged by apache2, which would have originated from 
DBI and DBD::Pg:

[Fri May 13 20:39:152011] -e: DBD::Pg::st execute failed: ERROR:  portal 
"" does not exist at /home/www/directory_to_module/a_module.pm line 514.

What I'm looking for is insight into what this error means, and help 
determing whether this is user error in our configuration or usage of 
PGPool2, or a potential bug in PGPool2. I'd like to note in case it 
helps, that when Load balancing is turned off, all portal errors go 
away. However, in our application we need loadbalancing as our 
application has reached a point where it peak traffic demands better 
database performance, and continues to grow)

I could post up our pgpool.conf, but it's nothing special, but if it 
helps let me know and I'll send it out.

Thanks for any help or insight!

-Jonathan

-- 
Jonathan Regeimbal
The Richard Group
w: 703.584.5808
c: 540.907.5116
e: jlregeim at gmail.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://pgfoundry.org/pipermail/pgpool-general/attachments/20110514/619c034c/attachment.html>